(My outline of the steps.)

1. There are 3 kinds of signs for association: icons which depend on contiguity or closeness, indexes which depend on similarity or correlation, and symbols which are defined by arbitrary convention (e.g. the word “table” for a certain piece of furniture). Contiguity and similarity are also the two kinds of magic, as described by Frazer in “The Golden Bough”.
   Indexical relations are involved in conditioned response kind of learning, which can be done with both animals and humans, when reinforcement of the associations to be learned is achieved by rewards (food) and punishments (electric shocks). The animal essentially learns the persistence of a correlation between the token and the object. Symbolic learning is different: it is learning by insight, as in Gestalt psychology (seeing the configuration as a whole, the forest not the trees). These two kinds of learning illustrate the difference between knowledge and understanding.
   However, a word standing by convention for a thing (as a noun does) or an activity (as a verb does) or a quality (as an adjective does) is still only an indexical relation, according to Deacon. Symbolic reference comes in only when relations between symbols (horizontally) are recognized and habitually used. This involves abstraction, classification, and generalization. He claims that this is highly counterintuitive.
2. Deacon claims that the symbolic transition is very difficult to achieve for animals thinking in iconic and indexical terms, because it involves some unlearning of old habits as well as the learning of new ones, thinking in a new mode. It seems that only humans have achieved this transition, although some apes can catch on after strenuous training, if they are still young and plastic.
   The transition to symbolic reference in humans is associated with the beginnings of language, some 1.5 million years ago, in Homo habilis and then Homo erectus. But it is not clear if symbolic reference led to language or vice versa.
3. The transition is not directly related to brain size, either absolute or relative to body size. Rather, it is related to a reorganization of brain circuits. Language (or any other brain function) is not localized in any particular brain area, but is distributed to many areas through neuron connections 2nd circuits. The axons can be very long and reach across areas, as is already evident in embryonic development when these connections are made. An excess of neurons are formed, and those that do not reach their proper targets are eliminated by apoptosis (programmed cell death).
4. The prefrontal cortex gained predominance over circuits previously used for other functions. Circuits normally compete for access, in a quasi-Darwinian fashion, and dominance (a “larger vote”) is gained by greater use. (Cf. Dennett’s “demons” in “Consciousness Explained”). Then the prefrontal cortex also grew in relative size.
   While ape brains increased in comparison to average mammalian brains, relatively, because bodies got smaller, human brains increased relatively in comparison with ape brains because brains got bigger for the same body size.
   Reorganization of the brain involves both neuron proliferation in some parts (e.g. the prefrontal cortex and some other areas) and different parcellation, i.e. how many neuronal connections are received in and sent by particular brain structures. I.e. a re-wiring is involved.
5. Use of language created a bigger and more differentiated brain, not vice versa. But they mutually supported each other (coevolution). The mechanism was Baldwinian evolution: behavioural changes affect the customs of society (social or cultural evolution by “memes”); then society exerts selection pressure on individuals (e.g. less fluent language speakers are less desirable as mates and thus leave fewer descendants); therefore the genetic propensity for fluent language skills increases. Behaviour affecting genes is almost Lamarckian, but not really; the mechanism is strictly Darwinian.
6. Language use was advantageous in society because it was needed for social contracts: (a) for peace, to counteract Prisoner’s Dilemma situations and promote reciprocitytype cooperation, as in Axelrod’s “Evolution of Cooperation”, and (b) for marriage, to reconcile permanet pairbonding needed for child-care, with larger group cooperation among men for hunti~g large animals.
7. Hunting and use of stone tools for butchering became necessary when meat eating had to supplement the original vegetarian diet due to shortages in the ice age. Men had to be away for prolonged times, since women with small children could not hunt. Marital fidelity had to be assured by means of a binding contract: the pair bond, involving promises of both fidelity and provisioning, had to be established by a publicly witnessed contract, with penalties for betrayal. Rituals were invented for this purpose. Both the contract and the ritual required language.
8. Pair-bonding replaced polygyny when fathers were needed to bring food to dependent mothers and children. Child care came to require more work and time as infants were born earlier and more helpless. This in turn was because the heads got too big to go through the birth canal, because of the growth in brain size. So what caused what? It seems to be a cycle.
9. Physical changes other than brain reorganization were also needed for efficient speech production, especially descent of the larynx in the throat. There was a cost for this: a greater probability of choking while eating, especially when talking at the same time.
10. The habit of symbolic reference and the use of language led to a different type of consciousness. Animals using icons and indexes have a consciousness, but it is different in kind. We still have this kind of animal consciousness in the background. It is non-verbal, consisting of sensations, emotions, and impulses. It is studied in a philosophical tradition called phenomenology. In any case, Whorf was wrong when he said that we cannot think at all without words.

Humans can form the idea of the self (as children do at some stage), and also recognize a similar self in other humans. (Some apes can recognize their own self, but not that of others.)

This makes humans into responsible moral agents. They can deal with others on a nI-Thoun basis. Unfortunately, this has not yet completely taken hold in human behaviour. By mentally dehumanizing others, we can justify mistreating them. We still do this on a widespread world-wide scale, but increasingly we hold war criminals in the Light – at least in the light of public (and perhaps soon legal) disapproval. (I am referring to the imminent creation of a World Criminal Court. )

In any case, world society is still evolving. Perhaps this further transition to becoming fully responsible moral agents is still in progress.